- OpenAI Dots always-on agent is a persistent AI assistant powered by GPT-6 Astra, announced at DevDay on September 29, 2026.
- Each Dot runs on a dedicated cloud computer with its own browser for long-running tasks.
- Dots connect to 4,000+ apps and follow up through ChatGPT, email, Slack, and messaging channels.
- Unlike a normal chat, a Dot keeps working toward a persistent goal between your conversations.
- Configure permissions and autonomy settings to control exactly what each Dot can access.
What Is the OpenAI Dots Always-On Agent?
The OpenAI Dots always-on agent is OpenAI's persistent AI assistant product, introduced at DevDay on September 29, 2026. Instead of relying on one-off chat sessions, each Dot is a continuously running agent driven by GPT-6 Astra, OpenAI's flagship model for complex reasoning, computer use, and research tasks. Every Dot gets its own cloud computer and browser, allowing it to connect tools, execute multi-step work over hours or days, and report back through ChatGPT, text message, email, or Slack.
The core distinction is persistence. A standard ChatGPT conversation ends when you close the tab. A Dot keeps its goal, its files, its connected apps, and its task history alive, so work continues even when you are away from the keyboard.
| Feature | Normal ChatGPT Chat | OpenAI Dot |
|---|---|---|
| Session lifetime | Ends when conversation ends | Always-on, persists between sessions |
| Execution environment | Your current session | Dedicated cloud computer + browser |
| Task type | Single-turn or short multi-turn | Long-running, scheduled, recurring |
| Follow-up channel | In-chat only | ChatGPT, email, Slack, messaging |
| Underlying model | Varies by plan | GPT-6 Astra |
Use a Dot when a task benefits from continuity: ongoing research, recurring monitoring, scheduled reports, or cross-app workflows. For a quick question, a regular chat is still the right tool.
OpenAI Dots Setup: Step-by-Step
Creating your first Dot takes just a few minutes. The setup flow lives inside ChatGPT on desktop web or the supported desktop experience, and it is built around defining a clear goal rather than writing a single prompt.
Start a New Dot
Open the Dots experience in ChatGPT on desktop web or the desktop app and choose to create a new dot. Access depends on your plan and the current rollout.
Choose a Name
Give the Dot a short, descriptive name tied to its responsibility, such as a project, workflow, or research topic.
Define Its Goal
Write the ongoing outcome the Dot should work toward. A focused goal gives the Dot stable direction across multiple work sessions.
Set Working Instructions and Autonomy
Describe what the Dot should prioritize, what boundaries it must follow, and how independently it should act before asking for approval.
Attach Context and Files
Add relevant documents and reference material so the Dot has the project context it needs when executing tasks.
Assign the First Task and Review Activity
Give the Dot a concrete task connected to its goal, then track progress from its activity dashboard, where active, scheduled, and completed work appears.
Dots access is plan-dependent and rollout is staged. Mobile access varies by account and platform, while creation and advanced configuration are centered on desktop experiences. Check your workspace settings if you do not see the Dots option.
Cloud Computer, Apps, and Always-On Work
Two capabilities set the always-on agent apart: the dedicated cloud computer and the connected-app ecosystem.
The cloud computer gives each Dot its own remote environment with a browser. Work there continues independently of your local machine, so a Dot can complete multi-step browser tasks while you do something else. You can open the computer view to inspect its state, and step in manually whenever review or intervention is useful. Cloud execution is separate from optional local-computer access, keeping remote agent work distinct from anything touching your own device.
Connected apps expand what a Dot can reach. OpenAI's launch materials highlight 4,000+ app connections, plus communication channels like Slack and personal email for proactive follow-ups.
| Capability | What It Does | Best For |
|---|---|---|
| Dedicated cloud computer | Runs browser and multi-step tasks in the Dot's own environment | Long workflows, web research |
| Browser interaction | Operates web pages inside the cloud computer | Form filling, monitoring sites |
| Connected apps (4,000+) | Extends workflows across external services | Cross-app automation |
| Communication channels | Sends updates via ChatGPT, email, Slack | Proactive follow-ups |
| Scheduled tasks | Runs recurring or future-dated work | Reports, reminders, monitoring |
Ongoing Research
- Assign an evolving topic
- Dot keeps collecting and organizing findings
- Updates accumulate across sessions
Recurring Monitoring
- Repeatedly checks changing information
- Surfaces meaningful updates automatically
- Works between conversations
Cross-App Workflows
- Combines actions across permitted apps
- Handles email, calendar, and files
- Completes multi-service tasks end to end
Scheduled work turns a Dot into a true ongoing assistant: set a recurring check once, and the Dot delivers proactive updates without a new prompt each time. Review everything from the task management view.
Permissions, Autonomy, and Workspace Controls
Because an always-on agent acts on your behalf, permission boundaries matter. Dots use configurable controls at the user, workspace, and Enterprise levels.
| Control | Scope | What It Governs |
|---|---|---|
| Connected app permissions | User and workspace | Which apps and services the Dot may access |
| Cloud computer access | Dot capability | Whether the Dot can use browser-based computer tasks |
| Local computer access | Optional user capability | Separates your own device from the cloud environment |
| Autonomy settings | User configuration | How independently the Dot acts between interactions |
| Workspace availability | Administrator | Whether members can use Dots at all |
| Enterprise access controls | Enterprise admin | Organization-level governance and restrictions |
OpenAI has also added a dedicated Dots section to its official safety documentation, reflecting the product's agentic scope.
When configuring a new Dot, begin with narrow app permissions and lower autonomy. Expand access only after reviewing its activity dashboard and confirming the Dot handles tasks as expected.
In Business and Enterprise workspaces, administrators control whether Dots are available to members and which connected capabilities are permitted. Individual users cannot override workspace-level restrictions.
Your First Week with a Dot: Checklist
First Week Goals:
- Create your first Dot with a focused, single-goal definition
- Attach project files and context before assigning tasks
- Connect one app such as email or Slack
- Set up one scheduled recurring task
- Review the activity dashboard and adjust autonomy settings
- Verify connected-app permissions match your comfort level
FAQ
Q: What is the OpenAI Dots always-on agent?
Dots are persistent AI assistants powered by GPT-6 Astra, announced at DevDay on September 29, 2026. Each Dot has its own cloud computer and browser, connects to apps, and continues working toward assigned goals between your conversations, following up via ChatGPT, email, Slack, and other channels.
Q: How is a Dot different from a normal ChatGPT conversation?
A normal chat ends when the conversation ends. A Dot keeps its goal, context, files, and task history alive in a dedicated cloud environment, supports scheduled and recurring work, and can proactively send updates through connected communication channels.
Q: Where can I create and manage Dots?
Creation and configuration are centered on ChatGPT desktop web and the supported desktop experience. Mobile access varies by account and rollout stage. Business and Enterprise workspaces can manage availability through administrator controls.
Q: Can I control what my Dot is allowed to access?
Yes. Dots use configurable permissions covering connected apps, cloud and local computer access, and autonomy settings. Workspace administrators can additionally restrict whether members can use Dots and which capabilities are permitted.